plotScatterPerformance: Performance scatter-plot

In jakobbossek/asr: Algorithm Selection in R

Description

Generates a scatter-plot of pairwise performances of two solvers. The function expects a data frame x which contains performance values row-wise. The mandatory column “solver” encodes the name of the corresponding algorithm.

The procedure works as follows: the data frame x column is filtered to all rows where solver$x is either algo.a or algo.b. Next, a transformation of the data frame is conducted via dcast to produce a wide data frame with each one column for algo.a and algo.b respectively. These columns contain the performance values indicated by argument value.var.

Usage

plotScatterPerformance(x, value.var = "PAR10", algo.a, algo.b,
  color = "group", shape = "group")

Arguments

x	[data.frame] Data frame with columns at least “solver” and value.var.
value.var	[character(1)] Name of column which contains the performance values.
algo.a	[character(1)] Name of first algorithm (x-axis).
algo.b	[character(1)] Name of second algorithm (y-axis).
color	[character(1)] Name of column which determines coloring of points, e.g. instance group.
shape	[character(1)] Name of column which determines the shape of points, e.g. instance group.

Value

[ggplot] ggplot2 object.

jakobbossek/asr documentation built on April 25, 2020, 9:31 p.m.